Like Andersen’s Likelihood-Ratio test, the approach of Kubinger et al. (2009) shares the most frequently referenced assumption of the Rasch model – that is that the item difficulty parameters are statistically independent of the person ability parameters. The approach illustrated in this paper is now a three-way analysis of variance. It has offered a clearer testing of Rasch model.
Comments:
1. Further research may disclose which combination of item parameter interval and ability parameter’s standard deviation leads to analyses where the nominal Type I risk holds.
2. Different combinations of the number of items, the number of respondents, the relevant DIF, the Type I risk, and the power within the stated frame of restrictions could be the focuses of further studies.